#ml & #deeplearning can almost always find patterns in data
#ml & #deeplearning can almost always find patterns in data. But how often is this scientifically meaningful? We argue that a framework of strong hypotheses coupled with adversarial controls is a way to find out.
#MachineLearning & #AI are now widely used in #chemistry & #chembio. But, as with any science, it is still important to run the right controls. Our new In Focus piece from Keiser & Chuang @UCSF advise on adversarial controls to ensure models are meaningful https://t.co/rjd0zdLluX
— Laura Kiessling (@ChemicalBiology), October 19, 2018